Machine Learning Lab

Lab 11 - Support Vector Machine

---

S Shyam Sundaram
19BCE1560
October 18, 2021

Dr Abdul Quadir MD
L31+L32

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import plotly.express as px
%matplotlib inline

Generating non-linear separable data

---

from sklearn.datasets import make_circles

# generate 2d classification dataset
X, y = make_circles(n_samples=100, noise=0.1)

# scatter plot, dots colored by class value
df = pd.DataFrame(dict(X1=X[:,0], X2=X[:,1], Y=y))

colors = {0:'red', 1:'blue'}

fig, ax = plt.subplots()

grouped = df.groupby('Y')
for key, group in grouped:
    group.plot(ax=ax, kind='scatter', x='X1', y='X2', label=key, color=colors[key])
plt.show()

from sklearn.model_selection import train_test_split
X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=0.3)

Linear SVM

---

from sklearn.svm import SVC

linear=SVC(kernel='linear')
linear.fit(X_train,y_train)

SVC(kernel='linear')

from sklearn.metrics import accuracy_score

pred=linear.predict(X_test)
accuracy_score(y_test,pred)

0.4666666666666667

Polynomial SVM kernel

---

df

	X1	X2	Y
0	-0.755754	-0.701927	0
1	-0.324131	0.631230	1
2	0.512493	1.065001	0
3	0.156620	-0.766278	1
4	0.479098	0.667054	1
...	...	...	...
95	0.049305	1.025803	0
96	-1.148978	0.323755	0
97	-0.626174	0.605153	0
98	0.961855	0.403435	0
99	-0.379807	-0.473364	1

100 rows × 3 columns

df['X3']=df['X1']**2
df['X4']=df['X2']**2
df['X5']=df['X1']*df['X2']
df

	X1	X2	Y	X3	X4	X5
0	-0.755754	-0.701927	0	0.571164	0.492702	0.530485
1	-0.324131	0.631230	1	0.105061	0.398452	-0.204601
2	0.512493	1.065001	0	0.262649	1.134228	0.545806
3	0.156620	-0.766278	1	0.024530	0.587182	-0.120015
4	0.479098	0.667054	1	0.229535	0.444961	0.319584
...	...	...	...	...	...	...
95	0.049305	1.025803	0	0.002431	1.052271	0.050578
96	-1.148978	0.323755	0	1.320151	0.104818	-0.371988
97	-0.626174	0.605153	0	0.392094	0.366211	-0.378931
98	0.961855	0.403435	0	0.925165	0.162760	0.388046
99	-0.379807	-0.473364	1	0.144253	0.224074	0.179787

100 rows × 6 columns

X=df.drop(['Y'],axis=1)
y=df.Y

X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=0.25)

fig = px.scatter_3d(df,x='X1',y='X2',z='X5',color='Y')
fig.show()

fig = px.scatter_3d(df,x='X3',y='X4',z='X5',color='Y')
fig.show()

poly=SVC(kernel='poly')
poly.fit(X_train,y_train)

SVC(kernel='poly')

pred=poly.predict(X_test)
accuracy_score(y_test,pred)

0.76

RBF kernel

---

rbf=SVC(kernel='rbf')
rbf.fit(X_train,y_train)

SVC()

pred=rbf.predict(X_test)
accuracy_score(y_test,pred)

0.84

Sigmoid kernel

---

sig=SVC(kernel='sigmoid')
sig.fit(X_train,y_train)

SVC(kernel='sigmoid')

pred=sig.predict(X_test)
accuracy_score(y_test,pred)

0.68